On 10/5/24 11:14 PM, Eli Zaretskii wrote:
If we are seriously considering rewriting Texinfo in a different
language, why should we limit ourselves to C++? Nowadays there are
better, safer languages out there.
Which ones?
If we restrict ourselves to languages with fast execution, modest
run-time-size, ahead-of-time compilation, with an extensive utility library,
and a reasonable number of people who know the language, the only ones
I can think of are C++ and Rust. However, I have not been paying as much
attention
to programming languages as I used to.
I'm impressed by Rust, and it would probably be a good implementation language
for Texinfo.
However, it is not as mature as C++, has a steeper learning curve, and I think
fewer
"GNU people" know it well. (I have only written one fairly small program in
Rust,
to wrap the Tauri/Wry framework for DomTerm.) Plus there is currently only a
single
implementation of Rust - and it is not gcc-based (and thus not GNU). Gccrs, the
Rust
frontend to gcc, seems to be coming along fast, but it is a huge job.
C++ has another huge advantage over Rust in that is more-or-less a superset of
C,
so it is much easier to convert C to C++ than to Rust.
Of course rewriting the package is a very large job, and should not be
decided upon lightly.
However, converting existing C code to (working but not idiomatic) C++ is
pretty easy.
--
--Per Bothner
[email protected] http://per.bothner.com/
